Data formats

Creating dataframes

Making a df from files in folder

library(tidyverse)

#creating the dir_to_df function
dir_to_df <- function(files) {
  files <- gsub("^./","",files)
  df <- data.frame(column= unlist(files)) %>% 
    separate(column, sep="/", into=c("site","image"))
}
my_df <- dir_to_df(files)

Making a df from vectors

# vectors
employee <- c('John Doe','Peter Gynn','Jolie Hope') #character vector
salary <- c(21000, 23400, 26800) #numeric vector
startdate <- as.Date(c('2010-11-1','2008-3-25','2007-3-14')) #date vector
#combine into df
employ.data<-data.frame(employee,salary,startdate)
str(employ.data)

## 'data.frame':    3 obs. of  3 variables:
##  $ employee : chr  "John Doe" "Peter Gynn" "Jolie Hope"
##  $ salary   : num  21000 23400 26800
##  $ startdate: Date, format: "2010-11-01" "2008-03-25" ...

employ.data2<-data.frame(employee,salary,startdate,stringsAsFactors=TRUE) #employee names become factors

Dataframe manipulations

Accessing data

#like a list
employ.data["employee"] #returns a df

##     employee
## 1   John Doe
## 2 Peter Gynn
## 3 Jolie Hope

employ.data$employee #returns a vector

## [1] "John Doe"   "Peter Gynn" "Jolie Hope"

employ.data[["employee"]] #returns a vector

## [1] "John Doe"   "Peter Gynn" "Jolie Hope"

employ.data[[1]] #returns a vector

## [1] "John Doe"   "Peter Gynn" "Jolie Hope"

#like a matrix 
head(employ.data,n=2) #display two rows

##     employee salary  startdate
## 1   John Doe  21000 2010-11-01
## 2 Peter Gynn  23400 2008-03-25

employ.data[2:3,] #display row 2 & 3

##     employee salary  startdate
## 2 Peter Gynn  23400 2008-03-25
## 3 Jolie Hope  26800 2007-03-14

employ.data[employ.data$salary > 21000,]

##     employee salary  startdate
## 2 Peter Gynn  23400 2008-03-25
## 3 Jolie Hope  26800 2007-03-14

employ.data[1:2,2] #as vector

## [1] 21000 23400

employ.data[1:2,2, drop = FALSE] #as df

##   salary
## 1  21000
## 2  23400

Modifying 1 value

employ.data[1,"salary"] <- 30; employ.data

##     employee salary  startdate
## 1   John Doe     30 2010-11-01
## 2 Peter Gynn  23400 2008-03-25
## 3 Jolie Hope  26800 2007-03-14

Turn NAs into 0

mydata[is.na(mydata)] = 0

Adding columns/rows

employ.data2<-rbind(employ.data,list("Lea Katz",35000,"2021-10-01")) #adds a row
employ.data3<-cbind(employ.data2,age=c(35,45,27,26)) #adds a column
employ_data3<-employ.data2$age<-c(35,45,27,26) #another way to add a column

Removing columns/rows

employ.data3$age <- NULL #removes column
employ_data5 <- employ.data2[-1,] #removes first row

Change colnames/rownames

#baseR
"start" -> colnames(employ.data)[3]
#dplyr
iris %>% rename(petal_length = Petal.Length)
#from a csv... sometimes not well read
area<-header.true(area) #makes first column into title and removes the auto-generated ones

Creating subsets

library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

#filter with a condition
employ.subset<-subset(employ.data,subset = salary==21000 )
#select only some variables
employ.data %>% select(salary,employee)

##   salary   employee
## 1     30   John Doe
## 2  23400 Peter Gynn
## 3  26800 Jolie Hope

Find value and change it

data<-as.matrix(data)
m <- data[,-1]
rownames(m) <- data[,1]

for (x in 1:nrow(sum)) {
  i <- sum$image_filename[x]
  j <- sum$label_names[x]
  m[i,j]<-sum$cover[x]
}

Group by a variable and only keep last value

tlog<-aggregate(tlog,by=list(as.character(tlog$unix)), FUN=last)

Group by a variable and make a sum

# for 1 variable
sum<- area %>% 
  group_by(image_filename, label_names) %>% 
  summarize(cover = sum(cover))
# for all variables (if numeric)
summed<-mydata %>% 
  group_by(region.code) %>% 
  summarise(across(everything(),sum,na.rm=TRUE))

not sure what this does anymore…

names(images)<-as.character(unlist(images[1,]))

Create new columns with operations

library(dplyr)
employ.data %>% mutate(double_salary = salary*2)

##     employee salary  startdate double_salary
## 1   John Doe     30 2010-11-01            60
## 2 Peter Gynn  23400 2008-03-25         46800
## 3 Jolie Hope  26800 2007-03-14         53600

Merging 2 dataframes

Binding columns (rows match)

employ.info<-cbind(employ.data,personal.info)
#OR
employ.info<-bind_cols(employ.data,personal.info)

Matching values for common variables

mydata<-inner_join(mydata,metadata, by="transect")

Creating Lists

Making a list from df and vectors

age_vec<-c(35,45,27,26)
mylist<-list(employees = employ.data, ages = age_vec)

Vectors

Change “numerical vector” into “named numerical vector”

ages<-c(35,45,27,26)
names<-employ.data$employee
names(ages)<-names #gives names to numerical values of the vector

Simple things

Paste

#in a string
dive<-"20220601-DIV1"
campaign<-"zeeland"
my_path = paste0("data/",campaign,"/",dive,"/images")

#in a function 
file<-read.csv(paste0("tlog/csv-files/",dive,"-gps.csv"))

Plots

Spatial data

Position points

ggplot(graham_ind,aes(date,tree_cover)) +
  geom_point() +
  theme_light()

Tracked position

ggplot(graham_ind,aes(longitude,latitude)) +
  geom_path() #in order of dataset, use geom_line for ascending order

Full positioning plot with all the labels and titles

ggplot(graham_ind,aes(longitude,latitude)) +
  #geom_point() + # order is important, best if points are on top of path
  geom_path(color="grey67") +
  geom_point(color="pink") +
  geom_rug(sides="rb", alpha = .05) + #rb=right+bottom, alpha=transparency
  #xlab("Longitude") +
  #ylab("Latitude")  + #labels
  #ggtitle("Trajectory of Caribou C12") but "labs()" is better
  labs(
    x = "Longitude", #use NULL if you don't want any label
    y = "Latitude",
    title = "Trajectory of Caribou C12",
    subtitle = "Map of 2,562 locations of female C12 tracked between 2001 and 2003 in Graham, Canada",
    caption = "Data: B.C. Ministry of Environment & Climate Change",
    tag = "Fig. 1"
  )

Coordinate system

#linear and non-linear
ggplot(graham_ind,aes(longitude, latitude)) +
  geom_path() +
coord_cartesian(      
    ylim = c(56.8, NA) #you can set the min and max values of both axis
  )

#THIS WILL NOT REMOVE POINTS OUTSIDE THE PLOT AREA, just zoom
#if you want to really remove the points do like this :
ggplot(graham_ind,aes(longitude, latitude)) +
  geom_path() +
  scale_y_continuous(limits = c(56.8, NA))

## Warning: Removed 823 row(s) containing missing values (geom_path).

#another features
ggplot(graham_ind,aes(longitude, latitude)) +
  geom_point() +
  coord_cartesian(
    expand = FALSE, #to have à (0,0) origin
    clip = "off",         #so points/labels don't get cut on the axis lines
    ylim = c(56.8, NA) 
  )

#coord_fixed
ggplot(graham_ind,aes(longitude, latitude)) +
  geom_point() +
  coord_fixed(
    ratio = 5  #default is 1 if you put nothing
  ) +
  scale_y_continuous(
    breaks = seq(56,58, by = .04)
  )

#coord_polar : makes straight lines into curves! 
ggplot(graham_ind,aes(longitude, latitude)) +
  geom_path() +
  coord_polar()

Statistics

Boxplot

ggplot(graham_ind,aes(season,tree_cover)) +
  geom_boxplot() +
  theme_linedraw()

Assigning, saving, rendering, etc.

Assign plot to object

#g<-ggplot(graham_ind, aes(longitude,latitude)) +
  #geom_point()
#g
(g<-ggplot(graham_ind, aes(longitude,latitude)) +
    geom_point()) #to print it directly !! use parentheses
g +
  geom_path(color="purple") +
  geom_rug(side="rb",alpha= .08) # you can add layers to your plot object later... can aboid typos if you cant to change things?

Save

ggsave(filename="myggplot.png", # raster graphic
       width = 10, height = 7,
       dpi = 700) # pixel intensity to 600 or higher!

ggsave(filename="myggplot.pdf", #vector graphic => best
       width = 10, height = 7,
       device=cairo_pdf) # doesn't work?

Different types of plots

Discrete variable

ggplot(graham_ind,aes(
  longitude, latitude,
  color = season, #if color is a variable in the dataset, put inside "aes()"
  shape = season
  )) + 
  geom_point()

Continuous variable

ggplot(graham_ind,aes(
  longitude, latitude,
  color = tree_cover)) + 
  geom_point()

ggplot(graham_ind,aes(
  longitude, latitude,
  size = tree_cover
  )) + 
  geom_point(
    color = "darkgreen", #should not go in aesthetics because it is for the geom_points
    shape = 1 
  )

Scales

ggplot(graham_ind,aes(
  date, tree_cover,
  color = season)) + 
  geom_point() +
  scale_x_date() + #tells ggplot that the x axis is a date
  scale_y_continuous() +   #tells ggplot that the y axis is continuous
  scale_color_discrete()

ggplot(graham_ind,aes(
  date, tree_cover,
  color = season)) + 
  geom_point() +
  scale_x_date(
    expand = c(0,0), #removes the "padding" on the sides)
    date_breaks = "3 months", #can use english to specify date breaks
    date_labels = "%m/%y", #to shorten the dates to desired abbreviation
    name = NULL
  ) +
  scale_y_continuous(
    labels = scales::percent_format(scale=1), #our numbers were already 75, but if it was 0.75, we could do scale = 100
    sec.axis = dup_axis(name = NULL), #duplicate the y axis for easier reading
    name = "Tree Cover"
  ) +   
  scale_color_discrete(
    type = c("#AE79F0","#EEAF5C"), #either use normal color names or hexcodes
    name = "Season:"
  )

Color gradients

ggplot(graham_ind,
       aes(longitude,latitude,
           color = yday)) +
  geom_point () +
  scale_color_continuous (
    type = "viridis" # with existing palette, install more palette packages :)
  )

ggplot(graham_ind,
       aes(longitude,latitude,
           color = yday)) +
  geom_point () +
  scale_color_gradient (
    low = "#AE79F0", high = "#EEAF5C" # custom gradient
  )

heatmap

ggplot(graham_ind,aes(longitude,latitude,)) +
  geom_hex (
    aes(fill=..count..), # is useful to put inside the geom if there are several geoms
    bins = 15,
    color = "white", #white outlines draws your eye to global extremes
    size = 0.5 
  ) +
  scale_fill_continuous(type = "viridis") +
  scale_color_continuous(type = "viridis", guide = "none") +
  labs (
    x = "Longitude",
    y = "Latitude",
    title = "Heatmap of C12",
    caption = "Data source : blablabla"
  )

## Warning: Computation failed in `stat_binhex()`:

heatmap without the outlines

ggplot(graham_ind,aes(longitude,latitude,)) +
  geom_hex (aes(color =..count..), bins = 15) +
  scale_fill_continuous(type = "viridis",
                        name = "Count:",
                        #breaks = c(50,100,150,200,250)) 
                        #breaks = seq(50, 250, by = 50))
                        breaks = 1:5*50 #from 1 to 5 and multiply each by 50
                        ) +
  scale_color_continuous(type = "viridis", guide = "none") +
  labs (
    x = "Longitude",
    y = "Latitude",
    title = "Heatmap of C12",
    caption = "Data source : blablabla"
  )

## Warning: Computation failed in `stat_binhex()`:

#this is another way to do it  
ggplot(graham_ind,aes(longitude,latitude,)) +
  geom_hex (bins = 15, color = "white", size = 1) +  #bins = resolution => always try several
  rcartocolor::scale_fill_carto_c(palette="SunsetDark",
                                  name = "Observations:",
                                  breaks = 1:5*50)

## Warning: Computation failed in `stat_binhex()`:

Mutliple plots together (facets)

g <- ggplot(graham_ind,aes(longitude,latitude)) +
  geom_path(color="grey67") +
  geom_point(aes(color=factor(year))) + #aes here only colors the points!
  scale_color_discrete(
    name = "Year",
    #labels = c("year 2001", "year 2002", "year 2003")
    labels= function(x) paste("Year", x))
g + facet_wrap(~ season) #tilde ~ is option + N

g + facet_wrap(~factor(year)) #to add ylabs to mini-plots again => lemon package

to arrange the plots differently

g + 
  facet_wrap(
    ~ factor(year),
    ncol = 2
    )

#two pariables => GRID
g + 
  facet_grid(
    factor(year) ~ season
  )

Aestethics

Color palettes

display.brewer.all(type="seq") # "Sequential" : from light to dark
display.brewer.all(type="div") # "Diverging" : dark-light-dark
display.brewer.all(type="qual") # "Qualitative" : visual differences between groups
display.brewer.all(colorblindFriendly=TRUE)

Using a palette in ggplot

ggplot(graham_ind,aes(longitude,latitude,color=factor(year))) +
  geom_path(color="grey67") +
  geom_point() + #aes here only colors the points!
  scale_color_brewer(
    palette="YlOrRd",
    name = "Year",
    #labels = c("year 2001", "year 2002", "year 2003"),
    labels = function(x) paste("Year", x)
  )

Maps

Mapview

caribou <- st_as_sf(graham_ind, coords = c("longitude", "latitude"), crs = 4326, remove = FALSE)
mapview(caribou)

Useful Commands

Data formats

Creating dataframes

Making a df from files in folder

Making a df from vectors

Dataframe manipulations

Accessing data

Modifying 1 value

Turn NAs into 0

Adding columns/rows

Removing columns/rows

Change colnames/rownames

Creating subsets

Find value and change it

Group by a variable and only keep last value

Group by a variable and make a sum

not sure what this does anymore…

Create new columns with operations

Merging 2 dataframes

Binding columns (rows match)

Matching values for common variables

Creating Lists

Making a list from df and vectors

Vectors

Change “numerical vector” into “named numerical vector”

Simple things

Paste

Plots

Spatial data

Position points

Tracked position

Full positioning plot with all the labels and titles

Coordinate system

Statistics

Boxplot

Assigning, saving, rendering, etc.

Assign plot to object

Save

Different types of plots

Discrete variable

Continuous variable

Scales

Color gradients

heatmap

heatmap without the outlines

Mutliple plots together (facets)

to arrange the plots differently

Aestethics

Color palettes

Using a palette in ggplot

Maps

Mapview